Liquid Metal Leaching for Rare Earth Magnet Recycling

Abstract

This study investigates the optimization of liquid metal leaching for recycling rare earth elements (REEs) from NdFeB magnets, a critical step in addressing the increasing demand for these materials in various high-tech applications. We explored the effects of leaching time, stirring, and magnet demagnetization on the yield of the leaching process using molten magnesium. Conducted at 900 °C, our experiments assessed the leaching process over periods of 2, 3.5, and 5 h, with and without the application of stirring. Our findings show that longer leaching times considerably increase neodymium (Nd) and praseodymium (Pr) leaching yield, with a notable peak in efficiency found at 5 h. Stirring improved the uniformity of REEs significantly and resulted in up to 80% yield. Furthermore, our data show that pre-leaching magnet demagnetization improves leaching specificity, significantly reducing the presence of non-target metals like nickel and copper. These insights offer a pathway to more cost-effective recycling of REEs from magnet scrap, which is essential for environmentally conscious management of resources amid the escalating global demand for REEs.

1. Introduction

Rare earth elements (REEs) are widely used in a variety of industrial areas, including defense, transportation, information and communication technologies, and healthcare. According to the United States Department of Energy (DOE), the three most critical rare earth elements identified in the medium-term criticality matrix (2025–2035) are neodymium (Nd), praseodymium (Pr), and dysprosium (Dy). Based on the standards of the European Commission, each of these rare earth elements is a critical raw material (CRM), underscoring their strategic importance and scarcity [1,2]. Magnets using Nd, Pr, and Dy are utilized in renewable energy technologies for the manufacturing of generators of offshore wind turbines and traction motors of hybrid, electric, and hydrogen fuel cell vehicles. Electric vehicle traction motors nearly all use rare earth permanent magnets, and their market exceeded 10 million in 2022, which was 55% up relative to 2021. Within the United States alone, electric vehicle sales are expected to grow from 1.4 million in 2020 to 6.9 million vehicles by 2025 [3,4,5]. Over the last 11 years, more than 1.5 million plug-in hybrid electric cars (HEVs) and battery electric vehicles (BEVs) have been commercially produced worldwide. Permanent magnets are used in a vast number of electrified cars, and over 700 parts of a vehicle contain these rare earth magnets. Their application has been expanding significantly in electronics and clean energy technologies, such as actuators, motors, wind turbines, and electric vehicles, and this is projected to drive up demand in the future [6,7,8]

Permanent magnets, especially those containing rare-earth elements, are valued for their high maximum B-H product (BH_xam), often referred to as energy content, enabling the design of smaller yet more powerful magnetic motors, generators, and other applications. Among the various types of rare-earth magnets, samarium cobalt (SmCo) and neodymium iron boron (NdFeB) are the most prevalent. Notably, NdFeB magnets, often referred to as neo magnets, stand out for their exceptional BH_max and cost-effectiveness [9], as Nd is much more abundant than Sm and Fe is much more abundant than Co.

To enhance their durability and corrosion resistance, these NdFeB magnets are typically coated. The most widely used coating method is nickel and copper electroplating, which typically involves a three-layered structure consisting of nickel, copper, and then nickel again (Ni-Cu-Ni). NdFeB-based magnet materials have increasingly become predominant in the market for high-field permanent magnets. In recent years, the permanent magnet industry has leaned towards these magnets due to their superior properties, comprising approximately 30–35 wt% rare earth elements (REEs), 1 wt% boron, and 65 wt% iron [10,11]. The REEs primarily include Nd, Pr, Dy, and Tb [12,13].

Forecasts of rapidly increasing worldwide demand for Nd and other REEs indicate that recycling will need to become an important supply of these critical materials, augmenting increased mining and primary production. Though the relatively long lifetime and rapid growth rates of the above products will likely mean that recycling can only meet a small fraction of demand at this time, as these products mature, recycling can potentially provide a much larger fraction of demand. For this reason, many studies have looked into different extraction techniques over time to recover rare earth elements from NdFeB magnets, including solvent extraction [14,15], bioleaching [16,17], hydrogen decrepitation, called Hydrogen Processing of Magnetic Scrap (HPMS) [18,19], and liquid metal leaching [10,20].

The development of the liquid metal extraction technique was aimed at addressing the cost and energy use associated with aqueous processing and electroslag remelting technologies, as shown graphically in Figure 1. This “melt, leach, distill” magnet-to-metal approach involves using liquid magnesium to leach neodymium from NdFeB magnet scrap. It is based on the much higher solubility of rare earth metals such as Nd in liquid Mg vs. Fe and B [21]. In this process, the Nd is leached into the magnesium, forming a liquid Mg-Nd alloy. This liquid alloy can then be separated from the insoluble solid iron and boron particles [22]. Note that because it extracts rare earth elements as metals, this process does not require the costly and energy-intensive oxide separation and reduction steps of other magnet-to-oxide methods.

Nd can then be extracted from the Mg-Nd alloy using distillation, as the vapor pressures of Mg and Nd are many orders of magnitude apart. In the past, this has been costly and inefficient. A novel distillation method known as gravity-driven multiple effect thermal system (G-METS) can potentially reduce both the energy consumption and utility costs of Mg alloy distillation enough to reduce the overall Nd recycling flowsheet cost by up to 67% compared to traditional distillation methods, as reported by Chinwego et al. [23]. This is because G-METS reuses condensation energy to evaporate metal, reducing the energy consumption of most of the Mg removal by 80–90%, to less than 0.6 kWh/kg distilled Mg, compared with 5–7 kWh/kg for traditional Mg distillation [24,25].

As an alternative, Mg-Nd alloy with Nd content enriched up to about 30 wt% via G-METS distillation can be used as a master alloy for the Mg casting industry. Several Mg alloys incorporate up to 2 wt% Nd by weight, and casting operations typically blend such alloys from Mg-Nd master alloys rather than pure Nd metal.

For the majority of magnet and alloy applications, praseodymium and neodymium can be used together. The natural ratio of neodymium to praseodymium is roughly 3:1 in most rare earth ores, and since the Nd and Pr do not need to be separated, this results in cost savings. Additionally, the residual iron-boron material left behind from this process presents recycling opportunities, especially for low-grade iron castings where the exact composition is less critical [26,27].

Previous studies have shed light on the dynamics of neodymium diffusion in liquid magnesium. Xu et al. (2000) observed this process in NdFeB scrap of a specific size range of 20–50 mesh at temperatures up to 750 °C and leaching times between 2 and 8 h, deducing an average diffusivity of Nd in molten Mg to be approximately 8.98 × 10⁻⁸ cm²/s. Aligning with the findings of Chae et al. (2014), it is understood that increasing leaching temperatures accelerates the diffusivity of rare earth elements [10,28]. Consequently, a leaching temperature of 900 °C is expected to yield a higher concentration of Nd in the resultant Mg-Nd alloy.

Though lower in concentration, the heavy rare earths often found in magnets Dy and Tb command much higher unit prices and can have comparable or even higher values than the light rare earths Nd and Pr. For this reason, Ott et al. investigated the effectiveness of various leaching agents in separating them from rare earth magnets in two streams [29]. For example, liquid Mg can recover Nd and Pr from NdFeB scrap at high yield and Dy and Tb at low yield; then, a subsequent leaching step using liquid Bi can recover the remaining Dy and Tb [29,30]. Chinwego et al. modeled the cost of these operations, including the use of G-METS distillation [23] and summarized the materials flow graphically as shown in Figure 2.

The correlation between leaching time and rare earth concentration in the alloy has been a subject of debate. Na et al. (2014) concluded that an increase in leaching time correlates with higher rare earth content, while Sun et al. (2014) contended that the final Nd concentration in the Mg-Nd alloy is unaffected by variations in leaching time [11,31]. Okabe et al. showed leaching yields of 95–98% but with very long holding times of 24–72 h [32].

The authors have previously considered the effect of leaching time and magnet particle size on leaching yield [33]. However, that study did not consider the effect of stirring on rare earth yield, nor did it study the effect of demagnetization on the leaching of nickel and copper coatings.

The primary objective of this study is to explore the leaching behavior of NdFeB magnets in liquid Mg under various conditions with industrially realistic leaching times of up to 5 h. Key areas of investigation include the influence of different leaching times on the leaching process, the impact of stirring on the molten magnesium-rare earth (Mg-RE) alloy, and the role that nickel coating plays in leaching efficacy. In particular, partial oxidation of the nickel-rich coating dramatically reduces the nickel content of the product metal.

2. Materials and Methods

This research centered on extracting rare earth elements from NdFeB magnets, employing magnets from two different sources. The first set, from Radial Magnets (Radial Magnets Inc., Boca Raton, FL, USA), had a NiCuNi coating, while the second set, from BJA Magnetics (BJA Magnetics, Leominster, MA, USA), had a carbon coating. Despite their different coatings, the chemical composition of these magnets was similar: 71.4 wt% Fe, 21.7 wt% Nd, 5.70 wt% Pr, 1.0 wt% B, and 0.57 wt% Dy.

The process began by demagnetizing some of the magnets by heating them to 350 °C and holding them at that temperature for an hour in a Mellen furnace under an argon atmosphere, a step crucial for removing their magnetic properties. The magnets were informally tested for magnetic forces between them, and no such forces were found, indicating at least a 99% reduction in remnant magnetization. After demagnetization, the coatings on some of these magnets were carefully stripped off, while coatings were left on some. All of the magnets, whether demagnetized or not, whether stripped or not, were then mechanically crushed into fine particles of 80–100 mesh.

In the next stage, magnesium rods (99.8%, Strem Chemicals, Newburyport, MA, USA) were combined with the crushed magnet particles in a graphite crucible (5″ outer diameter, 4.5″ inner diameter, 5″ inner depth, fine extruded, GraphiteStore.com, Northbrook, IL, USA), maintaining a Mg:magnet mass ratio of 3:1. To prevent the oxidation and vaporization of magnesium, the furnace chamber was filled with argon gas to about 1 atm pressure before heating commenced. The temperature was set to 900 °C in order to achieve the maximum leaching rate with relatively little Mg evaporation.

Furthermore, the study explored the role of stirring in the process. We examined the effects of stirring on the Mg-Nd alloy at different leaching times (2–5 h). This stirring used a ⅛” (3 mm) diameter 440C stainless rod, bent so it swept out an arc through the charged metal and magnet scrap and turned about 180° through the crucible approximately 30 times per minute. This alloy consists only of elements with low solubility in Mg, particularly no Ni. The rod insertion point at the cool top of the chamber was sealed with an elastomer o-ring to minimize air ingress during rotation of the rod. Experiments with stirring featured about ten minutes of stirring in this way, followed by 20 min of idle time, and repeating ten minutes of stirring, etc., throughout the 2–5 h leaching duration.

After the completion of the leaching process, crucibles were allowed to cool by shutting off the furnace power, and the Mg-Nd alloy solidified within the crucible above the remaining residue which was mostly Fe-B. Once solidified, the alloy was carefully extracted from the crucible, followed by sectioning and polishing for a detailed analysis.

The analysis phase of the experiment employed scanning electron microscopy (SEM, JEOL, Tokyo, Japan), augmented with energy dispersive spectroscopy (EDS, Oxford Instruments, Abington, Oxfordshire, UK), to characterize the sample’s microstructure and elemental composition. Additionally, precise chemical composition analysis utilized inductively coupled plasma–optical emission spectrometry (ICP-OES, Horiba, Piscataway, NJ, USA). Samples were taken from the Mg-Nd alloy ingot in three places, labeled the bottom (center axis ~2 mm from Fe-B particles), middle (approximate centroid), and top (center axis ~4 m from the top surface).

3. Results

3.1. Effect of Leaching Time

For leaching experiments with no stirring and demagnetized NdFeB magnet scrap, the data shown in Figure 3 indicate a clear trend: recovery of Nd and Pr from the alloy improves significantly with increased leaching time between 2 and 5 h.

3.2. Effect of Stirring on Rare Earth Leaching

Figure 4 illustrates the microstructural variation in the Mg-Nd alloy, observed from the bottom to the top of the cross-sectional area, for an experiment with no stirring. The presence of rare earth metals, characterized by their higher atomic number, is indicated by the lighter regions in the images, while the darker areas represent the magnesium matrix. A notable observation is the much higher concentration of rare earth metals at the bottom of the molten magnesium as shown by EDS analysis. This segregation can be attributed to the higher density of neodymium (6.89 g/cm³) compared to that of magnesium (1.58 g/cm³) in their liquid states, leading to this stratification.

However, the introduction of stirring during the leaching process significantly alters this distribution. As demonstrated in Figure 5, stirring results in a more homogeneous distribution of rare earth metals throughout the liquid magnesium. Furthermore, Figure 6 highlights an enhancement in the leaching yield of rare earth metals due to stirring and in the uniformity of rare earth concentration in the Mg solvent. This indicates that mechanical agitation plays a critical role in both the distribution of rare earth elements within the alloy and the overall efficiency of the leaching process. The total recovery yield of the light rare earths Nd and Pr was approximately 80% when using stirring, as described above (Results subsection Effect of Stirring on Rare Earth Leaching). This result is consistent with the findings of Okade et al. at 900 °C at 72 h leaching time [32]. It was difficult to estimate rare earth recovery in the non-stirred samples due to the variation in Mg alloy leachate composition.

Increasing the leaching time to 5 h while stirring showed a more even spread of rare earth elements throughout the magnesium matrix, enhancing the leaching yield, similar to what was observed in Figure 6. This resulted in an increase in rare earth concentrations not only at the bottom of the alloy but also throughout the middle and top regions. When the alloy was not stirred, the rare earth elements accumulated primarily at the bottom, with a significant reduction in the middle and top layers, as shown in Figure 7.

Figure 8 shows a comparative analysis of the leaching rates for Nd and Pr under stirred conditions over leaching times of 2 and 5 h. The data show that increasing the leaching period increases the leaching rate of both rare earth elements. Notably, the stirring motion used in the leaching process considerably impacts this efficiency. Stirring at the 5 h mark not only ensures an even distribution of the components in the molten alloy but also adds to an accelerated leaching rate. This effect is more pronounced as compared to the 2 h leaching time, indicating that the combination of prolonged time and mechanical agitation again improves the leaching yield.

Table 1. ICP-OES characterization of the Mg alloy after leaching using as-received magnets.

	Mg (wt.%)	Fe (wt.%)	Nd (wt.%)	Pr (wt.%)	Dy (wt.%)	Ni (wt.%)	Cu (wt.%)
Top	99	0.01	0.33	0.12	0	0.26	0.06
Middle	97	0.03	2.3	0.73	0.01	0.42	0.15
Bottom	90	0.07	7	2.12	0.03	0.6	0.27

and

Table 2. ICP-OES characterization of the Mg alloy after leaching using demagnetized magnets.

	Mg (wt.%)	Fe (wt.%)	Nd (wt.%)	Pr (wt.%)	Dy (wt.%)	Ni (wt.%)	Cu (wt.%)
Top	99.8	0.01	0.1	0.02	0	0.03	0.02
Middle	98.16	0.02	1.6	0.12	0	0.06	0.05
Bottom	90.25	0.02	7.25	2.3	0.02	0.1	0.09

below show the effects of demagnetization on leaching effectiveness. According to the results in Table 2, the demagnetization process led to a reduction in the contaminants, with Ni showing a decline of roughly 80% to 88.5% and Cu exhibiting a consistent reduction of about 66.7% across the sample compared to the results in Table 1. The demagnetization process resulted in a decrease in neodymium (Nd) content by approximately 69.7% at the top and 30.4% in the middle, whereas the bottom section experienced an increase of 3.6%. Similarly, praseodymium (Pr) levels experienced a reduction of around 83.3% at the top and 83.6% in the middle but showed an 8.5% increase at the bottom. The dysprosium (Dy) concentration is not only lower than both Nd and Pr but also significantly lower than its expected share based on concentration. This is consistent with the findings of Ott et al. as described above [29].

4. Discussion

Our investigation into the demagnetization effect revealed that magnesium effectively extracts rare earth elements from both as-received and demagnetized NdFeB magnets, as evidenced by the data in Table 1 and Table 2, as others have shown as well. However, a discernible difference was observed in the levels of nickel and copper contaminants. The leaching of as-received magnets resulted in higher quantities of these impurities, suggesting that these non-target metals are also being dissolved, as shown in Table 1. On the other hand, the demagnetized magnets exhibited a more than 80% decrease in nickel and copper content. One hypothesis is that nickel activity is reduced due to surface passivation by oxidation during demagnetization. The results suggest that the demagnetization process may enhance the selectivity of the leaching agent towards rare earth elements, potentially by altering the surface properties of the magnet coatings and thereby minimizing the unwanted dissolution of coating materials. Given the labor-intensive nature of coating removal, this could potentially reduce the cost of preventing coating materials from contaminating recycled rare earth metals using this method.

Further investigations could confirm this mechanism by testing the effect of surface oxidation on nickel and copper dissolution in liquid Mg. It would also be interesting to test the effects of other magnet coating passivating agents and surface reactions beyond oxidation, such as sulfidation using S₂, sulfation using SO₂, or fluoridation using a gas such as a hydrofluorocarbon or hydrofluoroolefin, as sulfation and fluoridation are more effective at passivating the liquid Mg surface than oxidation.

5. Conclusions

This study systematically investigated the complicated dynamics of rare earth element leaching from NdFeB magnets, providing significant insights into optimizing this critical recycling process. The study of the influence of leaching time on leaching efficiency indicated a direct correlation: as the leaching time increased, so did the extraction rate of neodymium and praseodymium, with the most apparent recovery happening at a 5 h duration.

The role of mechanical stirring in the leaching process was found to be significant, especially at extended leaching times. Stirring for 2 and 5 h markedly enhanced the homogeneity of rare earth element distribution within the liquid magnesium alloy, resulting in an 80% recovery rate of the targeted elements. This underscores the importance of stirring as a pivotal factor in maximizing leaching efficiency.

Furthermore, the impact of magnet demagnetization before leaching was scrutinized. The findings indicate that demagnetization facilitates a more selective leaching process, effectively reducing the co-extraction of non-desired magnet coating elements such as nickel and copper. Our hypothesis for this improved selectivity is the effect of surface passivation by oxidation during demagnetization.

In conclusion, our findings advocate for prolonged leaching times and consistent stirring to improve the recovery rates of valuable rare earth elements. Additionally, the demagnetization of magnets emerges as a beneficial step in improving the selectivity of the leaching process. These insights can significantly contribute to the advancement of sustainable recycling practices for rare earth elements, which are vital for a wide range of modern technologies.